地震多属性分析技术在薄层砂体预测中的应用  

Thin sandstone prediction with seismic multi-attribute analysis

作　　者：蔡义峰[1] 熊婷[1] 姚卫江[1] 邢成智[1] 仲伟军[1] 贾春明[1] 

机构地区：[1]中国石油新疆油田公司勘探开发研究院地球物理研究所,新疆乌鲁木齐830013

出　　处：《石油地球物理勘探》2017年第A02期140-145,共6页Oil Geophysical Prospecting

基　　金：中石油"新疆大庆"重大专项(2012E-34-01)资助

摘　　要：为准确预测准噶尔盆地中拐凸起新发现油层的二叠系上乌尔禾组三段储层(厚度较薄、横向变化快,而单纯地震属性或反演难以分辨)在平面上的展布,提出了针对中拐凸起上乌尔禾组三段薄砂层预测的技术方法—利用GeoEast地震多属性模式识别技术,通过提取多种地震属性并根据地震属性与薄层砂体厚度的相关性进行优选,利用BP神经网络分析定量预测了薄层砂体的平面分布。研究发现上乌尔禾组三段具有"沟槽富砂"的特征,三口井的实际钻探证明预测的薄层砂体厚度与钻遇厚度比较吻合,误差均不超过1m。The Member 3 in Permian Wuerhe Formation in Zhongguai uplift,Junggar Basin is characterized by thin reservoir and lateral rapid variation.It is difficult to distinguish this kind of reservoirs with only seismic attributes or inversion.In order to accurately predict reservoirs in the area,we propose an approach based on tools provided by GeoEast.First we extract a variety of seismic attributes.Then we choose some seismic attributes which are related to sand body thickness.Finally with BP neural network we predict quantitatively the thin sand body distribution.It is found that the Member3 in Permian Wuerhe Formation are grooved sand bodies.Our prediction matches very well with three drilled wells.The difference of sand body thickness between the prediction and the drilling information is not more than 1 meter.

关 键 词：薄层砂体 多属性分析 储层预测 上乌尔禾组 BP神经网络 

分 类 号：P631[天文地球—地质矿产勘探]